In “On Referring” Peter Strawson pointed out that incomplete descriptions pose a problem for Russell’s analysis of definite descriptions. Howard Wettstein and Michael Devitt appealed to incomplete descriptions to argue, first, that Russell’s analysis of definite descriptions fails, and second, that Donnellan’s referential/attributive distinction has semantic bite. Stephen Neale has defended Russell’s analysis of definite descriptions against Wettstein’s and Devitt’s objections. In this paper, my aim is twofold. First, I rebut Neale’s objections to Wettstein’s and Devitt’s argument and argue that (...) Neale’s attempt to provide an account of referential descriptions within a Russellian framework fails. Second, contra Devitt and Wettstein, I argue that the problem posed by incomplete descriptions shows that Donnellan was right in believing that referential descriptions refer even if the definite description fits nothing. (shrink)

In “Is semantics possible?” Putnam connected two themes: the very possibility of semantics (as opposed to formal model theory) for natural languages and the proper semantic treatment of common nouns. Putnam observed that abstract semantic accounts are modeled on formal languages model theory: the substantial contribution is rules for logical connectives (given outside the models), whereas the lexicon (individual constants and predicates) is treated merely schematically by the models. This schematic treatment may be all that is needed for an account (...) of validity in a formal setting, but it does not help to understand how proper and common nouns function in reality (not in models). Putnam then initiated the empirical study of such nouns to indicate, (i), that the popular Frege-Carnap account of them as (“disguised” compound) predicates is empirically incorrect, and, (ii), that they offer a new paradigm for a naturalistic semantics of natural languages. We take Putnam’s program a couple of steps further. First, we investigate the semantics of common nouns and argue that they refer (to kinds), rather than apply by satisfaction/truth to a designation/denotation. Second, we point to general results about semantics as a theory whose fulcrum is the reference relation rather than satisfaction in models and validity across them. (shrink)